5A corrupts eMMC (was 5A does not boot)

#30

I wish I could delete posts in my thread. Y’all need to take this convo somewhere else, please.

Regarding htop, by default it doesn’t show the kernel threads, which apparently means dietpi’s htop isn’t showing the truth about your system, while arbian I guess tweaks htop to show kernel threads by default. Big f’ing whoop. Can y’all go bicker about this somewhere else? The bottom line is that yes, dietpi does install a more slim distro, and there’s a need for this from certain users. It’s all open source, you can see what the scripts are doing and there’s no secrets.

3 Likes
#31

You’d be hard-pressed to explain how the size matters. I bought this eMMC with the 5A from AllnetChina, not some random ebay merchant. You’d think it would work.
But you do confirm that the 5A and eMMCs does work, so that’s interesting.

#32

My 64GB eMMC module comes from Allnet as well.
I’ve just successfully made it work by switching from HS400 to HS200.
Basically, I had to apply this commit to rock-5a.dts:

diff --git a/arch/arm64/boot/dts/rockchip/rk3588s-rock-5a.dts b/arch/arm64/boot/dts/rockchip/rk3588s-rock-5a.dts
index 55fd83e8bc9c..63cdded53823 100644
--- a/arch/arm64/boot/dts/rockchip/rk3588s-rock-5a.dts
+++ b/arch/arm64/boot/dts/rockchip/rk3588s-rock-5a.dts
@@ -407,15 +407,13 @@ &saradc {
 /* Storage */
 /* eMMC */
 
 &sdhci {
        bus-width = <8>;
-       no-sdio;
-       no-sd;
        non-removable;
        max-frequency = <200000000>;
-       mmc-hs400-1_8v;
+       mmc-hs200-1_8v;
        /delete-property/ mmc-hs400-enhanced-strobe;
        pinctrl-names = "default";
        pinctrl-0 = <&emmc_rstnout &emmc_bus8 &emmc_clk &emmc_cmd &emmc_data_strobe>;
        status = "okay";
 };
#33

I also don’t understand why.

That is not stealing, but usually depends on the deal we have. If you took my software, you took that deal even you didn’t read it. Software is almost never provided without terms of use.

Taking the credits is.

Everyone is selling, even you. This story, your stand point. Money is not the only currency there is.

Exactly, there is intent to manipulate with you. Next question should be: why? But I think we already know the answer.

Fighting bad players is hard. Users don’t care, we know that too.

#34

Sorry. I won’t continue this.

2 Likes
#35

Can you try overriding and increasing the pull-up level (drive-strength) of all the data lines here: kernel/rk3588s-pinctrl.dtsi at stable-5.10-rock5 · radxa/kernel · GitHub ?

PCB traces may be a bit too long (or other issue) for the default settings, hence why it works fine in lower HS mode or inherently slower cards (like the 16 GB one).

#36

Sorry, I don’t really know how to do this. Can you give me some example of how it’s done?

#37

Similarly, I am having trouble booting Rock 5A from emmc.
After powering on, the blue LED does not light up or even if it does
dimmc_rockhip fe2c0000.mmc: busy; trying anyway
and cannot proceed to the next step.
I would like to know the solution.
The emmc and power supply are purchased from allnet china.

#38

Like what @GinKage said, I was not able to reproduce this on 16GB Foresee eMMC module. I recently worked around a different eMMC issue on ROCK 4 by reducing to HS200 so I’ll test if that solves the issue.

@dylandn @hiro and @nyanmisaka please leave your eMMC size and model here as well.

#39

One official Foresee 64G eMMC module from Allnet and one homebrew KIOXIA 64G eMMC module (THGAMRG9T23BAIL). Both of them work fine with the official USB 3 eMMC reader and Rock 5B.

#41

iIt’s a Samsung 16GB KLMAG1JETD-B041

#42

@RadxaYuntian Also confirmed that downgrading to HS200 fixes the issue on my 5A with the 64G eMMC modules. So I’m guessing there might be a compatibility issue with the board itself.

#43

Actually I think it was Foresee having issue at HS400. We reproduced the issue in our office, and I’ll try adjust pin drive as suggested above.

Still need to figure out why Samsung was not working for @dylandn though.

#44

I tried to increase it from level 2 to 5 but it didn’t help. Is it still safe to set this to higher values?

The maximum value of it is 15 - pcfg_pull_up_drv_level_15

#45

In which case I think drive strength is unlikely to be the issue here. I’ll validate HS200 internally and create a PR if that solves the issue.

#46

My module seems to be identical to dylandn’s, identified in dmesg as “CJTD4R 58.2 GiB”.
Update: scratch that, my module has different capacity, I’ll post the part number later today.

#47

Can you take a picture of the front and back of your ROCK 5A and eMMC module?

I want to know the hardware version of ROCK 5A and the brand and capacity of eMMC module.

#48

Module info here: 5A corrupts eMMC (was 5A does not boot)

#49

I use the following emmc bought from allnet china.
EMMC 5.1 FOR ROCK 4 / 3A / 5B / 5A / E (ALSO FOR ODROID) 32GB

product page

emmc writer uses the following emmc writer purchased from allnet china
EMMC TO USD BOARD
product page

#50

Does your eMMC chip has FORESEE engraving?